Hi,
I'm running a 2 node cluster (VCS3.5 MP2) with vxvm 3.5MP2
and vxfs 3.5MP2). there is 2 service-groups. "a" and "b".
Their running 3 filesytems in each, with an application and informix
database. (vx informix database vcs-agent)
One of the filesystems in each servicegroup is running VSM 4.5MP5)(hsm)
and the nodes themselves is mediaservers (NB 4.5 MP5).
All solaris of course (sol8).....
We had an incident where one of the migrating filesystems was
running full, because new data arrived faster than VSM could
copy out to tape and purge filesystem.
then we had to do a 'migdbcheck -r'
Now, this migration wont work. We can do a manual migration/purge:
migrate <filename>
migrc -R <name>
migpurge <filename>
and it works fine.
When we let the scheduler see that we are past the watermarks,
migsweep gets the following error when it should start to copy out:
09/04 14:06:33 [9512]migcopy[9663]: ERROR: No source for 66BF40MBE61F9
09/04 14:06:33 [9512]migcopy[9663]: perror 22: Invalid argument
09/04 14:06:33 [9512]migcopy[9663]: REPORT: 66BF40MBE61F9 0 206 2 0 0 0 0
09/04 14:06:33 [9512]migcopy[9663]: ERROR: Failed to copy 0xBE61F9; tret = 2
09/04 14:06:33 [9512]migcopy[9663]: ERROR: copy_for_method() ret=2
Once for each file.....
Anyone seen this before?
The case is escalated to veritas, but we are a bit short of time here, so
if anyone have any hints, I'l send them a picture of some flowers!
Yours
--------------------------------------------------------------
Espen Martinsen Email : esm AT mnemonic DOT no
Mnemonic AS, Waldemar Thranesgt. 77 Phone : +47 2299 9710/00
Mobil : +47 915 88 485
N-0175 Oslo, NORWAY Fax : +47 2299 9701
--------------------------------------------------------------
(PADI AOW 87191) http://www.mnemonic.no/
pgp print = 99CD 4027 36F4 0A4F 68DA 2925 327A 5BCF 9C21 F277
|